Automatic refinement of ephemeris‐derived range‐Doppler model through the least‐squares matching of Sentinel‐1 Single‐Look Complex images and DEM in mountainous areas

Abstract

Range‐Doppler model is a transformation that relates ground and Single‐Look Complex SAR image spaces by knowing the instantaneous position and velocity of the sensor in its trajectory. Its instantaneous position and velocities are determined with the help of vectorial functions, which their precisions directly influence the accuracy of the range‐Doppler model. Ephemeris data is one of the sources used for the estimation of temporal functions. Some reasons can cause uncertainty in the temporal function directly derived from the satellite navigation data. Ground control points are other alternatives for calibration of the range‐Doppler model whose visual measurements are difficult in the mountainous areas of Sentinel‐1 Single‐Look Complex images due to their resolutions and distortions. The temporal functions of the range‐Doppler model are refined through numerous control points extracted from the automatic matching between Single‐Look Complex images and Digital Elevation Model. For this purpose, the radiometric and geometric similarities of the Single‐Look Complex images and Digital Elevation Model in mountainous areas are first increased based on the initial range‐Doppler model, and then their geometric discrepancies are compensated with the least‐squares image matching method. The results indicated that the proposed method in mountainous areas could automatically compensate for up to 30 pixels errors and reach sub‐pixel accuracies.